[PDF] Challenge and Retention in Games

View - Download Challenge and Retention in Games

Previous PDF

Next PDF

UNIVERSITY OF CALIFORNIA,

IRVINE

Challenge and Retention in Games

DISSERTATION

submitted in partial satisfaction of the requirements for the degree of

DOCTOR OF PHILOSOPHY

in Informatics by

Thomas Debeauvais

Dissertation Committee:

Professor Cristina V. Lopes, Chair

Professor Gary Olson

Assistant Professor Joshua Tanenbaum

2016

Parts of Chapters 3, 4, 5, and 7

c

2010-2016 ACM

All other materials

c

2016 Thomas Debeauvais

TABLE OF CONTENTS

Page

LIST OF FIGURES vi

LIST OF TABLES viii

ACKNOWLEDGMENTS x

CURRICULUM VITAE xi

ABSTRACT OF THE DISSERTATION xii

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Thesis and Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.4 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.5 Organization of the Dissertation . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 Related Work 8

2.1 Enjoyment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.1.1 Motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.1.2 Player Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.2 Retention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.2.1 Engagement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.2.2 Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.2.3 Longitudinal Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.3 In-Game Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.3.1 Social Sciences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.3.2 Improving Gameplay . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.3.3 In-Game Purchases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3 Ragnarok Online 22

3.1 Gameplay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.2 Private Servers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.3 Methods and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

ii

3.4 Supporting Group Play . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.4.1 Tweaking Group Parameters . . . . . . . . . . . . . . . . . . . . . . . 29

3.4.2 The who Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.4.3 Warpra and Healer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.5 Maintaining an Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.5.1 Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.5.2 The autotrade Command . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.5.3 The whosell Command . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3.6 Improving the Quality of Life . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.6.1 Character Customization . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.6.2 Control Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.6.3 Bot Hunting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.6.4 Community Management . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.6.5 Free-to-Play vs Pay-to-Win . . . . . . . . . . . . . . . . . . . . . . . 38

3.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4 World of Warcraft 40

4.1 Gameplay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.2 Methods and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.2.1 Dataset 1: First Survey . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.2.2 Commitment Metrics and Their Limitations . . . . . . . . . . . . . . 44

4.2.3 Gold Buying Metric and its Limitations . . . . . . . . . . . . . . . . 48

4.2.4 Dataset 2: Second Survey and Gameplay Data . . . . . . . . . . . . . 49

4.3 Commitment Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

4.3.1 Region and Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

4.3.2 Hardcore-Casual Dichotomy . . . . . . . . . . . . . . . . . . . . . . . 53

4.3.3 Play Motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.3.4 Guild Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.3.5 Playing With One's Partner . . . . . . . . . . . . . . . . . . . . . . . 55

4.4 Gold Buying Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4.4.1 Demographics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

4.4.2 Play Motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4.4.3 Links With Commitment . . . . . . . . . . . . . . . . . . . . . . . . . 59

4.4.4 Generalized Linear Model . . . . . . . . . . . . . . . . . . . . . . . . 60

4.5 Progression and Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.5.1 Segments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.5.2 Raid progression and replays over time . . . . . . . . . . . . . . . . . 68

4.5.3 Explaining presence and churn . . . . . . . . . . . . . . . . . . . . . . 72

4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

5 Forza Motorsport 4 78

5.1 Gameplay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

5.1.1 Cars and Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

5.1.2 Assists and Bundles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

5.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

iii

5.2.1 Race and Achievement Datasets . . . . . . . . . . . . . . . . . . . . . 83

5.2.2 General Trends and Distributions . . . . . . . . . . . . . . . . . . . . 84

5.3 Patterns of Assists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

5.3.1 First-race Bundle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

5.3.2 Most Frequently Used Bundles . . . . . . . . . . . . . . . . . . . . . . 85

5.3.3 Assist Progression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

5.3.4 Assist Transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

5.4 Modeling Assist Transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

5.4.1 Scenario 1: Predicting the Success of Disabling . . . . . . . . . . . . 90

5.4.2 Scenario 2: Characterizing a Successful Disabling . . . . . . . . . . . 92

5.4.3 Evaluating the Models . . . . . . . . . . . . . . . . . . . . . . . . . . 93

5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

6 Jelly Splash 96

6.1 Gameplay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

6.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

6.2.1 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

6.2.2 Data Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

6.3 Diculty Spikes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

6.3.1 Level Diculty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

6.3.2 Level Hopelessness . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

6.3.3 XMRs and Board Rerolls . . . . . . . . . . . . . . . . . . . . . . . . . 104

6.4 Chapter Gates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

6.4.1 Retention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

6.4.2 Purchases and Spending . . . . . . . . . . . . . . . . . . . . . . . . . 107

6.4.3 Facebook Logins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

6.5 Energy/lives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

6.5.1 Sub-Sampling and Mixed-Eects Regression . . . . . . . . . . . . . . 111

6.5.2 Interpreting Churning . . . . . . . . . . . . . . . . . . . . . . . . . . 112

6.5.3 Interpreting Purchasing . . . . . . . . . . . . . . . . . . . . . . . . . 113

6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

7 Undisclosed Mobile Game 115

7.1 Gameplay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

7.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

7.2.1 Telemetry Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

7.2.2 Dening Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

7.3 Pre-Tutorial Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

7.3.1 Inadequate Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

7.3.2 Longer Loading Times . . . . . . . . . . . . . . . . . . . . . . . . . . 121

7.4 Tutorial Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

7.4.1 Step by Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

7.4.2 Churn Over Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

7.5 Modeling Tutorial Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

7.5.1 Filtering and Cleaning-up . . . . . . . . . . . . . . . . . . . . . . . . 127

iv

7.5.2 Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

7.6 Post-Tutorial Churn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

7.6.1 Diculty and Hopelessness . . . . . . . . . . . . . . . . . . . . . . . . 132

7.6.2 Other Session Variables . . . . . . . . . . . . . . . . . . . . . . . . . 134

7.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

8 Lessons Learned 138

8.1 Challenge and Retention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

8.1.1 Tolerance to Diculty . . . . . . . . . . . . . . . . . . . . . . . . . . 139

8.1.2 Grind with Moderation . . . . . . . . . . . . . . . . . . . . . . . . . . 139

8.1.3 Player Perception and Feedback . . . . . . . . . . . . . . . . . . . . . 140

8.1.4 Adjusting Challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

8.2 Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

8.2.1 Learning Curve and Pacing . . . . . . . . . . . . . . . . . . . . . . . 141

8.2.2 Empirical Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

8.2.3 Early or Never . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

8.3 Segments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

8.3.1 The Cheetahs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

8.3.2 The Rhinoceroses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

8.3.3 The Butter

ies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

8.3.4 The Elephants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

8.4 Social Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

8.4.1 Boosting Retention . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

8.4.2 Hell is Other People . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

8.5 Money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

8.6 Miscellaneous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

8.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

9 Conclusion 149

Bibliography 152

A Compilation of Findings 163

v

LIST OF FIGURES

Page

3.1 Screenshot of a player shop in Ragnarok Online . . . . . . . . . . . . . . . . 24

3.2 Warpra and Healer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.3 Control Panel of a Ragnarok Online private server. . . . . . . . . . . . . . . 37

4.1 World of Warcraft screenshot. . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2 Distribution of Commitment Metrics in WoW . . . . . . . . . . . . . . . . . 46

4.3 Average WoW weekly play time for the three motivation factors. . . . . . . . 54

4.4 Ratio of gold buyers against the three motivation scores and across genders. 59

4.5 Ratio of gold buyers against duration of breaks, across regions and genders. . 60

4.6 Correlation network between variables from the rst WoW survey. . . . . . . 61

4.7 Ratio of active players, split by region and gender. . . . . . . . . . . . . . . . 67

4.8 Ratio of active players, split by \having taken a break before November 2011". 68

4.9 Active player base defeating each boss in Dragon Soul, split by region. . . . . 70

4.10 Active player base defeating each boss in the Firelands, split by region. . . . 71

4.11 Month-to-month transfer behavior of the World of Warcraft player base. . . 74

5.1 Mapping of the controls on an Xbox 360 controller. . . . . . . . . . . . . . . 82

5.2 FM4 achievements tracking player progression over time. . . . . . . . . . . . 84

5.3 Ratio of players enabling an assist over their races in career and online modes. 87

5.4 Scenarios of success, failure, and yoyo after disabling an assist for the rst time. 89

5.5 Rate of success, failure, and yoyo after disabling an assist for the rst time. . 89

6.1 Screenshots of Jelly Splash. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

6.2 Ratio of active players X days after install. . . . . . . . . . . . . . . . . . . 101

6.3 Diculty and churn for levels 10-30 in Jelly Splash. . . . . . . . . . . . . . . 103

6.4 Distribution of spending for free players and small, medium, and big buyers. 107

6.5 Percentage of players who login with Facebook for each level. . . . . . . . . . 109

6.6 Histogram of sessions, split by number of lives at session start and end. . . . 110

7.1 Percentage of users still playing against time since install. . . . . . . . . . . . 118

7.2 Diagram of install-related steps happening before the tutorial. . . . . . . . . 121

7.3 Percentage of installs that churned against median start time. . . . . . . . . 122

7.4 Dashboard graph of the players reaching each tutorial sub-step. . . . . . . . 125

7.5 Linear regression of churn against median step duration, with residuals. . . . 126

7.6 Correlation network between variables from the undisclosed game. . . . . . 130

vi

7.7 Diculty and churn for levels 50-70 in the undisclosed game. . . . . . . . . . 133

7.8 Hopelessness and churn for levels 50-70 in the undisclosed game. . . . . . . . 134

8.1 Shape of the retention curves for World of Warcraft, Forza 4, and Jelly Splash.142

vii

LIST OF TABLES

Page

1.1 Summary of the dissertation data sources. . . . . . . . . . . . . . . . . . . . 6

1.2 Listing of challenge and retention metrics used in the dissertation. . . . . . . 6

3.1 In-game issues and their solutions found on private servers. . . . . . . . . . . 28

3.2 Comparison of character customization on ocial vs private servers. . . . . . 36

3.3 Summary of ndings for RO. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.1 Partial World of Warcraft timeline. . . . . . . . . . . . . . . . . . . . . . . . 41

4.2 Variables collected in the second WoW dataset . . . . . . . . . . . . . . . . . 51

4.3 Average retention metrics for dierent segments of WoW players. . . . . . . 52

4.4 Average features and ratio of gold buyers for dierent WoW segments . . . . 57

4.5 Logistic regression model explaining gold buying. . . . . . . . . . . . . . . . 64

4.6 Stats of the average player over 7 months, split by region and gender. . . . . 66

4.7 Log odds ratios from regression models explaining presence and churn. . . . 73

4.8 Precision and recall of predicting monthly presence and churn. . . . . . . . . 75

4.9 Summary of ndings for WoW. . . . . . . . . . . . . . . . . . . . . . . . . . 77

5.1 Conguration of assists in built-in bundles. . . . . . . . . . . . . . . . . . . . 81

5.2 Frequency of bundles across modes. . . . . . . . . . . . . . . . . . . . . . . . 86

5.3 Sign of the log-odds ratios of a logistic regression in scenario 1. . . . . . . . . 91

5.4 Sign of the log-odds ratios of a logistic regression in scenario 2. . . . . . . . . 93

5.5 Precision and recall for each assist in both scenarios. . . . . . . . . . . . . . 93

5.6 Summary of ndings for FM4. . . . . . . . . . . . . . . . . . . . . . . . . . . 95

6.1 Common F2P monetization mechanics and their focus. . . . . . . . . . . . . 97

6.2 Linear regressions explaining churn and purchases in JS. . . . . . . . . . . . 106

6.3 Level-40 gate behavior. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

6.4 Sign of the log-odds ratios from two mixed-eects logistic regressions. . . . . 112

6.5 Summary of ndings for Jelly Splash. . . . . . . . . . . . . . . . . . . . . . . 114

7.1 Structure of the main dataset. . . . . . . . . . . . . . . . . . . . . . . . . . . 117

7.2 Percentage of players reaching each beginner milestone. . . . . . . . . . . . . 119

7.3 Number of installs, models, churn, and start time across versions. . . . . . . 119

7.4 Breakdown of tutorial steps. . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

7.5 Filtering criteria, and ratio of dataset excluded by each. . . . . . . . . . . . . 128

7.6 Odds ratio and signicance of factors explaining day-0 churn. . . . . . . . . . 129

viii

7.7 Performance of models predicting churn on day of install. . . . . . . . . . . . 131

7.8 Sign of the log-odds ratios from a logistic regression predicting churn. . . . . 135

7.9 Summary of ndings for the undisclosed game. . . . . . . . . . . . . . . . . . 137

8.1 Retention personas accommodating various ways to segment players. . . . . 143

A.1 Summary of ndings related to challenge and money. . . . . . . . . . . . . . 163 A.2 Summary of ndings related to segments, time, social, and miscellaneous. . . 164 ix

ACKNOWLEDGMENTS

I would like to thank the National Science Foundation for partial support during my doctoral program. Even though not part of this thesis, I enjoyed working with my advisor and my colleagues on other projects funded by the NSF. x

CURRICULUM VITAE

Thomas Debeauvais

EDUCATION

Doctor of Philosophy in Information and Computer Science 2016

University nameCity, State

Master of Science in Information and Computer Science 2012

Another university nameCity, State

xi

ABSTRACT OF THE DISSERTATION

Challenge and Retention in Games

By

Thomas Debeauvais

Doctor of Philosophy in Informatics

University of California, Irvine, 2016

Professor Cristina V. Lopes, Chair

Game designers and researchers agree that the main motivation for starting playing a game is challenge. It is only a small step to say that when the game becomes too dicult, players can become frustrated and quit. While extensive work shows that challenge is central in player enjoyment, its in uence on player retention has received little attention. Establishing this in uence is dicult for several reasons: (i) Denitions of challenge tend to be game- or genre-specic. (ii) Measuring retention accurately requires real-life data, which is often hard to access. (iii) Challenge can be intertwined with other factors, such as social interactions in multi-player games. Addressing these three limitations requires a mixed-method, holistic, and cross-game approach paying close attention to the game mechanics. Therefore, we use a diverse mix of data sources, ranging from interviews to online questionnaires and telemetry logs, to explore player behavior in ve games: Ragnarok Online, World of Warcraft, Forza Motorsport 4, Jelly Splash, and an undisclosed commercial mobile game. This work provides empirical evidence that challenge does in uence retention, and that many contextual factors, namely player segments, social interactions, time, and money, nuance and mediate this in uence. These ndings may help game developers make better games, expand our knowledge of challenge, retention, and their relationship, and improve the bottom-line of game companies by making games more engaging and players stay longer. xii

Chapter 1

Introduction

The past decade has seen the rise of three concurrent practices in the game industry. First, games user research, the branch of user experience research applied to games, has matured and is now standard among large game developers and publishers. Games user researchers study what players enjoy or dislike in a game. Their tools range from qualitative focus groups and interviews to quantitative surveys and telemetry data analysis [67]. A second phenomenon that has seen its way into the game industry is big data. More and more data is collected about how people play and what they do inside games. The data collected is also becoming increasingly complex and diverse: from eye tracking to diary studies, A-B tests, and physiological data, all triangulating player enjoyment [135]. A third phenomenon is the recent major shift in the typical game's genre, audience, lifecycle, platform, and business model. The 1990's and 2000's were the (second) golden era for video game consoles [47]. At the time, most games were released once and for all, in a re-and-forget manner

1. Game

prices were somewhat standard, and no game was given to players for free. By 2009, Facebook had reached a quarter billion users, oering a new platform for game developers to tap into a new and more casual audience [73]. By the early 2010s, social games like FarmVille were1 Exceptions include long-lasting franchises such as Madden Football or Final Fantasy. 1 played by tens of millions of players daily [85]. Successful games are now expected to start new or continue existing franchises [115] and release new content through regular patches.

1.1 Motivation

In this context, there has been an increased focus on player retention, i.e. how long players engage with the game, and when and why they drop out (or churn, as it is also called). Studying retention and its counterpart, churn, is interesting and valuable because they are a proxy for player enjoyment: when players like the game, they keep playing; when they stop liking it, they stop playing and churn [110]. Game designers are interested in how they can build more engaging games with better retention. One knob in their control is the game's challenge level [3, 111]. Best practices recommend making games that are easy to understand, progressively more dicult, and ultimately hard to master [98]. The assumption is that if the game is too dicult, players may become frustrated and churn. If the game is too easy, they may become bored and also churn [80]. This assumption is reasonable, but does it always hold empirically? After all, some players enjoy games that are incredibly dicult [74]. Others have fun playing games that are some- times trivially simple just to pass time or socialize [73]. And although designers know better than players what is best in a game, they sometimes make mistakes [29]. They also have certain tastes, and tend to only design and play games that t their tastes [80]. So designers may not always understand the appeal of certain game mechanics on certain segments of players [17]. That is where academic games user research can help study niches of players with particular tastes and expectations [37, 92, 101, 121]. While these academic studies focus on the motivations of players, they ignore or at least stay vague on game design particulars. 2 In the end, the relationship between challenge and retention does not seem straightfor- ward. Improving our understanding of this relationship may help game designers build better games, expand our knowledge of challenge, retention, and the in uence of one on the other, and improve the bottom-line of game companies by making players stay longer.

1.2 Thesis and Research Questions

In this thesis, I claim thatchallenge in

uences retention, but this in uence is highly contex- tual.The goal of this thesis is to empirically explore the in uence of challenge on retention by highlighting contexts and situations in which this in uence is manifest or surprisingly missing. In the process of conducting this work, I focus on the following research questions: (1) Is the in uence of challenge on retention observed empirically? As will be shown in the rest of the dissertation, the rst-order approximation is yes. However, dening challenge and churn is not trivial, and heavily game-specic. Once adequate metrics of challenge and churn are found, correlating them is more straightforward. (2) What factors mediate the in uence of challenge on retention? To answer this question both broadly and in depth, I gather the results of dierent methodological approaches applied to several games of various genres. Consistent with a wide range of literature covered in Chapter 2, I nd that social interactions, time, money, and player segments often mediate, nuance, and sometimes even eclipse, the in uence of challenge on retention.

1.3 Approach

My work comprises seven studies using a total of three dierent techniques on ve dierent games. By applying dierent techniques to study these games, I am able to nd more 3 ways in which challenge and retention relate, and I am able to identify more factors and contexts mediating this relationship. In each study, I place a strong emphasis on comparing the empirical behavior of players to the intended purpose of a game's mechanics. In the remaining of this section, I list the ve games investigated in my work. Each of the games listed below is covered in its own chapter later in this thesis. The rst study, detailed in Chapter 3, concerns private servers of a massively multiplayer online game (MMO) called Ragnarok Online (RO). Although these private servers are illegal, thousands of players would rather play on them than on the ocial servers. Do these players prefer private servers because they are easier? Through qualitative interviews and participant observation, I nd that diculty plays a role on player enjoyment and retention, but so do social interactions and the money that players are ready to spend into the game. I also provide game mechanics introduced exclusively on private servers to alleviate, if not completely address, the problems most commonly raised by players on ocial servers. The second, third, and fourth studies are combined in Chapter 4 and relate to World of Warcraft (WOW), another MMO. As of 2016, WOW remains one of the most popular MMOs, and a lot of academic research covers it [55, 91, 92, 96, 100, 133]. With 10 million players in 2012 [1], WOW gathered players from around the world, across genders, and of all ages [109, 131]. Do dierent demographic segments play and churn dierently? Challenge in WOW revolves around grinding for game gold, but players sometimes (illegally) buy game gold with real money. Which segments of players buy gold most? In MMOs, the lack of new content is also said to be the main cause of churn [89]. How fast do players consume new content once it is released? I answer these questions by analyzing the data from two online questionnaires answered by several thousand WOW players (to be able to segment players demographically), as well as in-game player behavior data collected over seven months (to measure actual churn, and not just the breaks that players report in the questionnaire). The fth study looks at progression and how players adapt to the game's challenge over 4 time in a racing game called Forza Motorsport 4 (F4). This study, presented in Chapter 5, is particularly relevant to my thesis because in the game, players can manually congure the amount of challenge they are ready to face. As for the approach, I analyze millions of logs from telemetry data collected over two years. Large-scale data mining is a very appropriate technique to study retention because it measures churn objectively, reliably, and unobtrusively [53]. The sixth study, covered in Chapter 6, focuses on the in uence of challenge on churn in the mobile tile-matching game Jelly Splash (JS). Its main question is the core of this thesis: does diculty make players churn? The approach consists, again, in mining millions of logs from telemetry data for the reasons mentioned above. This study also introduces a process and metrics for quantifying challenge and players' perception of challenge from in-game telemetry data in level-based games. The seventh and last study, detailed in Chapter 7, centers on the tutorial of an undisclosed commercial mobile game (U) installed by millions of players. Since the literature reports that tutorials can in uence long-term retention [5, 28], the goal of this study is to observe player churn before and during the tutorial. By mining data from a dozen versions released throughout the game's soft launch, I am able to post-hoc A-B test dierent aspects of the game and see how they in uence retention. I nd that time, and not a particular game mechanic, can explain nearly 90% of the churn happening during the tutorial. These seven studies result in a long and heterogeneous list of empirical ndings, graphs, behaviors, and player segments related to challenge, retention, or both. To aggregate and summarize items from this list, I build an anity diagram. The goal is to surface practical lessons learned that may be applicable to a wide range of games. These lessons learned are presented in Chapter 8. 5

GameNMethods

Ragnarok Online (RO)9interviews

30+participant observation

World of Warcraft (WOW)2,865online questionnaire

1,350online questionnaire paired with game data

Forza Motorsport 4 (F4)220,000+telemetry data

Jelly Splash (JS)274,000+telemetry data

Undisclosed game (U)78,000+telemetry data

Table 1.1: Summary of the dissertation data sources.

GameChallengeChallenge MetricRetention Metric

ROAcquiring equipmentPlayer reportsPlayer reports

Boss monsters

WOWBoss monstersBosses killedWeekly play time, breaks

Amassing goldBuying goldWeekly play time, breaks

F4DrivingDisabling assistsNA

quotesdbs_dbs27.pdfusesText_33

[PDF] aion serveur padmarashka

[PDF] aion 5.0 patch note

[PDF] camping donnacona

[PDF] camping un air d'été 2005

[PDF] un air d'été paroles

[PDF] roulotte a vendre camping un air d'été

[PDF] qu'est ce qu'un air d'opéra

[PDF] camping un air d'eté

[PDF] 459 route grand capsa

[PDF] pont-rouge

[PDF] qc g3h 1l3

[PDF] air d'opéra definition

[PDF] technique de lancer de poids

[PDF] lancer du poids exercices

[PDF] dimension du cercle de lancer de poids